Ferromagnetic molecular charge-transfer complexes - Chemical

Synthesis and Properties of the Heterospin (S1 = S2 =/2) Radical-Ion Salt .... Suranjan Shil , Satadal Paul , and Anirban Misra. The Journal of Physic...
0 downloads 0 Views 4MB Size
Chem. Rev. 1988, 88, 201-220

20 1

Ferromagnetic Molecular Charge-Transfer Complexes JOEL S. MILLER" Central Research and Development Department, E. I. du Pont de Nemours and Co., Inc., Experimental Station, €328, Wilmington, Delaware 19898

ARTHUR J. EPSTEIN' Department of phvsics and Department of Chemistry, The Ohio State University, Columbus, Ohio 43210

WILLIAM M. REIFF Department of Chemistry, Northeastern Universiw, Boston, Massachusetts 02 115 Recelved June 11, 1987 (Revised Manuscript Received October 8, 1987)

Contents I. 11. III . IV.

V.

VI. VII. VIII. IX.

Overview Introduction Magnetic Phenomena Magnetic Properties of [Fe(C,Me,),]'+[Anion]'A. Anisotropic Magnetic Properties B. Mossbauer Spectroscopy C. Structure-Function Relationship 1. Alternate Ligand Substituents 2. Alternate Anions 3. Alternate Metals Models for Ferromagnetic Coupling in Molecular Solids A. Heitler-London Spin Exchange B. High Spin Multiplicity Molecules and Polymers C. Configurational Mixing of an S = 1 Excited State To Stabilize Ferromagnetic Coupling 1. Homospin Systems 2. Heterospin Systems D. Generallzed Hubbard Model E. General Considerations F. Stabilization of Bulk Ferromagnetism Organic Ferromagnets Summary Acknowledgment References

20 1 202 202 204 209 209 210 210 210 21 1 21 1 211 21 1 213 214 215 216 216 216 217 217 218 219

I . Overview Some ionic molecular solids comprised of linear chains of alternating S = l/z metallocenium donors (D) and S = 1/2 planar polycyano hydrocarbon acceptors (A), i.e., .-D*+A.-D*+A*--.,exhibit cooperative magnetic phenomena, i.e., ferro-, antiferro-, ferri-, and metamagnetism. The high-temperature (7'1 50 K) magnetic Contribution No. 4431 from the Central &search and Development Department.

susceptibility can usually be fit by the Curie-Weiss expression, x cc (T with 0 > 0 for those salts with dominant ferromagnetic interactions and 8 < 0 for those salts with dominant antiferromagnetic interactions. For [Fem(C5Me5),]*+[TCNE]'-, magnetization in the absence of an applied field, Le., spontaneous magnetization or bulk ferromagnetic behavior, is observed below the Curie temperature of 4.8 K and the magnetization exhibits a hysteresis curve with a coercive field of 1 kG. Critical exponents p, y, and 6 characteristic of bulk ferromagnetic behavior are also observed. The 57FeM h b a u e r spectra show that this class of charge-transfer complexes exhibits zero applied field Zeeman-split spectra with large internal fields of 400-450 kG. In order to understand the magnetic coupling in this class of materials, a model based on the configurational admixture of the lowest charge-transfer excited state with the ground state was developed. Ferromagnetic stabilization is consistent with forward ( A e D ) , but not retro (D+A), charge transfer between the cation and anion for this class of compounds. The model is expanded to other electron configurations that contain singly, doubly, or triply degenerate partially occupied molecular orbitals (POMO) and offers a convenient guide to explore ferro-, antiferro-, and ferrimagnetic phenomena in molecular systems. Assuming the virtual charge transfer (retro or forward) excitation involves only the highest energy POMO, stabilization of ferromagnetic coupling via this mechanism requires that the stable radicals possess a non-half-filled degenerate POMO and the excited state possesses the same spin multiplicity and mixes with the ground state. Thus, the radical must possess a POMO with accidental or intrinsic (i.e., belonging to the DZd,Cs, or higher point group) orbital degeneracies. Since ferromagnetism is a bulk phenomenon, ferromagnetic coupling must be present and dominate throughout the solid. A model for achieving this is discussed. Novel materials as well as more intensive experimental data and deeper theoretical insight are necessary to test these concepts and establish a salient understanding of cooperative phenomena in molecular solids. The report of bulk ferromagnetism in [Fe111(C,Me5)2]'+[TCNE]'affords the opportunity to test these concepts in real systems.

0009-2665/88/0788-0207$06.5010 0 1988 American Chemical Society

202 Chemical Reviews. 1988. Vol. 88. NO. 1

Joel S.Miller was born in Detroit, MI, and received his Bachelor of Science in Chemistry from Wayne State Unhrerslty In 1967 and Ph.D. from UCLA in 1971. After a postdoctoral fellowship at Stanford University, he joined the Xerox Webster Research Center in 1972 and later joined the Occidental Research Corp. (ORC) in 1978. I n 1981 he was Visiting Professor of Chemistry at the University of California. Irvine, CA. and Chairman of the Southem California Section of the American Chemical Society. After the demise of ORC he joined the Central Research and Development Department at the Du Pont Co.. where he is currently a Research Supervisor for solld-state chemistry. His research interests focus On the solid-state electrical. magnetic, and optical properties of molecular (organic. organometallic. and inorganic coordination) compounds and charge-transfer complexes as well as the surface mdification of solids.

Arthur J. Epstein was born in Brooklyn, NY. and received h k B.S. in mysics from the Polytechnic Institute of Brooklyn in 1966 and his Ph.D. In Physics from the University of Pennsylvania in 1971. After 13 years as a scientist at the Xerox Webster Research Center. he joined The Ohio State University in 1985 as Professor of Physics and Professor of Chemistry. He is also adjunct hofessa of Physics at the Unhrersity of Fbrlda and has been a Vkflng Professor at the Technion-Israel Instiiute of Technology and the Unbersity of Park. His research interests Include synmetii metals. molecular magnetism. conducting polymers. and ceramic superconductors.

I I . Introduction With the first reports' by Du Pont researchers that segregated linear-chain, I-D, charge-transfer salts based upon TCNQ2 may exhibit high dc electrical conductivity as well as other novel solid-state propertiess and the conjecture that a high-temperature excitonic superconductor based upon a linear-chain compound might be prepared: the past quarter of a century has witnessed substantial multidisciplinary efforts focused

Miller et ai.

William M. Reiff was born in Binghamton. NY, and went on to receive his A.B. at S.U.N.Y. Binghamton in 1964 and his m.D. at Syracuse University in 1968. He was an NIH-PHS m u a e t felow from 1966 to 1968 and an associate scientist at Brookhaven National Labwatoly. He was an NSF faculty associate-posMoctoraI fellow in the Departments of Chemistry and Physics at the University of Texas. Austin, from 1968 to 1970. He joined the Department of Chemistry at NortheastemUniversity in 1970 and has been full professor since 1977. He is an active visiting scientist at the Francis B i e r National Magnet Laboratory. His current research interests include low dimensional magnetkm chargetransfer complexes. ultra-low-temperature magnetism of ionic iron fluorlde compounds, Mossbauer studies of gold compounds. metal ion insertion chemistry. and heterogsnwes catalysis via lowvabnt metal clusters.

toward understanding and exploiting this fascinating clans of molecular solids." In contrast, molecular solids possesing linear chains comprised of altemating donors (D) and acceptors (A), i.e., .-DADADA.-, are in general poor conductors; however, they exhibit a variety of interesting optical properties,' and recently special examples have been characterized to exhibit a temperature/presure-dependent change in the degree of charge transfer, i.e., a 'neutral-ionic" transition: as well as cooperative magnetic phenomena: e.g., bulk meta-9J0 and ferromagnetismg*"as well as a spin-Pierels transition.12 Herein after reviewing magnetic phenomena of molecular materials, we discuss the structurefunction relationship of an unusual c h of molecular solids which exhibit ferromagnetic behavior. Finally, models consistent with the stabilization of ferromagnetic coupling and bulk ferromagnetic behavior for this class of compounds are discussed.

I I I . Magnet/c Phenomena All substances exhibit a magnetic moment, M,upon application of a magnetic field, H,which is related to H by M = xH, where x is the magnetic susceptibility.l3*lrlas Open-shell paramagnetic compounds have their induced moment aligned parallel to the field. For noninteracting independent spins the magnetic moment is inversely proportional to temperature (T) and the susceptibility can be modeled by the Curie expression (eq 1)where x = g p f l / k B T or the more general Bril-

x

c = Ng2jcB2s(s

= C/T

1 ) / 3 k ~ T= (0.375 (emu K)/mol)(S(S

(1)

+ 1)g2)/3T

Chemical Reviews, 1988, Vol. 88, No. 1 203

Ferromagnetic Charge-Transfer Complexes

Ferromagnet Spontaneous Mbgnetization M(H.0) r 0

.".ti-

0

lw e = -10

Y

0

100

p/",

~

( 2 J2J + 1)x ) - $ c t n h ( & )

louin (eq 2) expressions, where N is Avogadro's number, g is the Lande factor, pB is the Bohr magneton, J = S L , x gJpBB/kBT, and kB is the Boltzmann con~ t a n t . l ~Closed-shell ~-~ diamagnetic compounds have their induced moment aligned antiparallel to the field. This latter phenomenon is temperature independent. In some circumstances these spins experience an effective parallel (or antiparallel) exchange (molecular or Weiss) field due to the neighboring spins which leads to an increase (or decrease) in the measured susceptibility from that predicted far independent spins. The high-temperature susceptibility data often may be expressed by the Curie-Weiss14 law (eq 3), where for ferromagnetic (parallel) and antiferromagnetic (antiparallel) interactions 8 is respectively greater or less than zero. x = C/(T - e) (3)

+

peff =

I

I

/

I

Diamagnetic Field, H, G

Figure 2. Schematic illustration of the magnetization (M) as a function of field (If)for several types of commonly observed magnetic behavior.

C

50 Temperature .T, K

Figure 1. Susceptibility (x) (a), reciprocal susceptibility (x-l) extrapolated from the high-temperature region (b), and effective moment (fi ) (c) as a function of temperature for independent g = 2, S = spins as well as ferromagnetically coupled (e = 10 K) and antiferromagnetically coupled (e = -10 K) spins.

Bj(x) = , + ' c t n h ( 2J

/

(3xk~/NT)l/~ = 2.823(xT)'I2 = p B k 2 s ( s + 1)1112(4)

The magnitude of x is temperature dependent and chemists frequently report the effective moment,'" peff (eq 4). The susceptibility,reciprocal susceptibility (x-l), and effective moment possess characteristic temperature dependencies. The magnetization ( M ) also possesses a characteristic field dependency, which enables the rapid qualitative determination of the magnetic behavior. These dependencies are illustrated for in-

dependent spin (Curie), ferromagnetic, and antiferromagnetic behaviors in Figures 1 and 2. At sufficiently low temperature the spins may order. If they align parallel to each other (ferromagnet), then a macroscopic spontaneous magnetization at zero applied field [Le., M(Happ= 0) > 01 is present with a characteristic saturation moment, M , (e.g., 1.22 X lo4 (emu G)/mol for Fe),Iadin a finite applied field. The saturation magnetization, Ma,can be calculated from eq 5. If neighboring spins are aligned antiparallel Ma = NgspB (5) (antiferromagnet), then there is no net macroscopic moment in zero applied field and the susceptibility is anisotropic below the Nee1 temperature. Ferrimagnetism occurs when the antiferromagnetically aligned spins have differing local moments resulting in incomplete cancellation of the parallel and antiparallel spin sublattices leading to a reduced, but nonzero, moment. The saturation magnetization for a ferrimagnet can be calculated from eq 6 or 7 depending if incomplete

Ma = N&SpB

(6)

M , = NgASpB (7) cancellation of sublattice magnetic moments arises from differences in g or S, respectively. Application of a magnetic field to a ferromagnet leads to alignment of the ferromagnetic domains and M(H) exhibits a hysteresis behavior with a characteristic coercive field (e.g., -1 G for Fe)laEnecessary to move the domain walls. Metamagnetism is the field-dependent transformation from an antiferromagnetic state to a high-moment ferromagnetic state. Like the gas/liquid critical behavior, the onset of cooperative magnetic behavior near the Curie temperature, T,, can be scaled with critical exponents; i.e., the phenomena can be modeled by (T - Tc)x,where X is the critical exponent.15 The critical exponents can be compared against theoretical expectations to elucidate the dimensionality and anisotropy of the dominant spin interactions. The highly magnetic ferro- (e.g., Fe, CrOz) and ferrimagnetic (e.g., Fe,O,) materials frequently find utility in technological applications ranging from household

Miller et al.

204 Chemical Revlews. 1988, Vol. 88. No. 1

TABLE 1. Slunmary of W e t i c Phenomena Found in Molecular Compounds" magnetic magnetic effect of field and field gradient susceptibility, susceptibility magnetic behavior A300 K): emulmol x(30 K ): emu/mol product, H d H / d z diamagnetism --lo+ --lo+ repulsive paramagnetism' 1.25 X IOJ 12.5 X IOJ attractive antiferromagnet&"* 1.19 x 10J 8.33 X loJ attractive ferromagnetismb* 1.32 X loJ 25.0 x 10-3 attractive ferrimagnetisme >+lo4 complex attractive metamagnetismc >+10-3 eamplex attractive

x(T) none 0:

molecular examde

T'

eamplelP complexe complex complex

'X M/H.'Assuming one S = 'I2,g = 2 unparied electron per repeat unit (Assuming 0 = 15 K. *At higher temperature -(T- e)-' with €> I 0. 'At higher temperature =(T- e)-' with 0 < 0.

..

ventional synthetic organic chemistry as well as the ease of fabrication enjoyed by organic materials may lead to their use in future generations of electronic and/or photonic applications.

I V. Magnetic Properties ot [Fe(C&leJJ'+[Anlon] *Our strategy for designing molecular ferromagnets evolved from research aimed at preparing TCNQ2 (1)-based sublimable conducting charge-transfer salts

.. Figure 3. Alternating donor/aeceptor, -D'+ A-D'+A--, line. ar-chain structural motif observed for Fem(C&lea)2~*IA]" (A = TCNQ,TCNE, DDQ, CdCN&), IF (C~H,)ZIITCNEI, lFem~CsMe6~zl'+IC,~CN~sI~, and IM (C,Medzl+ITCNEl'~(M = Co, Ni). The structure shown here is for A = TCNE (open circles = H, closed circles = C, small hatched circles = N, and large hatched circles = M).

magnets to complex digital rmrdihg media and are the subject of contemporary materials research to improve their properties.16 Table 1summarizes the more common magnetic behaviors. Highly magnetic materials, i.e., ferro- or ferrimagnetic, have close approach of the spin-containing moieties (i.e., covalent- or metalliebonded 1-D, 2-D, or 3-D network structures) where the magnetic sites are d or f orbital transition, lanthanide, or actinide metal based. Examples of highly magnetic molecular-based c ~ m p o u n d s ' ~with J ~ p or even d orbitals contributing to the cooperative magnetic properties were until recently unknown. Their discovery parallels the quest for molecular-based superconductors" and the exploration of cooperative phenomena in quasi-1-D materials! This broad range of phenomena combined with the anticipated modulation of the physical properties via con-

containing organometallic cations. Two 1:l chargetransfer salts of [Fem(C,Me~21"[TCNQl'103Lcomposition have been prepared. The kinetic phase is comprised of a 1-D structural motif based on alternating S = 1/z[Fem(C,Me,)2]" (2) cation donors (D) and S = '/,[TCNQ]'anion acceptors (A), Le., D'+A'-D'+A--,21 as illustrated for the TCNE2 (3) salt (vide infra) (Figure 3). The effective moment (flSff) reveals a substantial deviationlo from the temperature-independent Curie behavior characteristic offerrocenium saltsz2 (Figure 4). The field-dependent magnetization reveals antiferromagnetic behavior (with an antiferromagneticordering temperature, TNd 2.55 K) for H < 1.6 kG for the [TCNQ]'. salt. In contrast, for H > 1.6 kG a sharp rise and approach to magnetization saturation characteristic of ferromagnetic behavior are observed (Figure 5).9J0 This field-dependent switching from antiferromagnetic to a high-moment behavior is consistent with metamagnetism; several examples of which, e.g., FeCI,, have been characterized.z3 This molecular chargetransfer complex, however, is the first example where neither a 1-, 2-, nor 3-D covalently bonded network structure is present-only a conventional organic solvent soluble molecular solid.17JR With this discovery of a molecular metamagnet we sought to prepare a molecular ferromagnet. The strategy was to elucidate the structurefunction relationship through modification of the radical anion, cyclopentadiene ring substituent group, and the metal to identify the combination of steric and electronic features leading to stabilization of ferromagnetic coupling as well as bulk ferromagnetic behavior. The primary

..-

-

Chemical Reviews, 1988, Vol. 88, No. 1 205

Ferromagnetic Charge-Transfer Complexes

.

i

.

*.

.*

I

(.

3

100

0

200

300 100

Temperature, T, K

200

300

Temperature, T, K

Figure 4. Effective moment (4 as a function of temperature fo; [Fem(C5Me&&+[A]- [A*,= TCNQ, TCNE, DDQ, C4(CN)6, C3(CN)5]and [M (C5Me5),J [TCNE]'- (M = Co, Cr, Ni). Data are for polycrystalline samples. r

0

f eoO0I

Figure 6. Reciprocal m y e t i c susceptibility (x-l) as a function of temperature for [Fen (C5Me&]'+[A]- [A = TCNQ, TCNE, DDQ,C,(CN),,C,(CN),] and [M (C5Me5)J+[TCNE]'- (M = Co, Cr, Ni). Data are for polycrystalline samples.

I

5000

10000

15000

20000

FIOld, H, 0

Figure 5. Magnetization (M) as a function of applied field (H) for [Fe"1(C,Me5)z]'+[A]'- [A = TCNQ, TCNE, C4(CN)6]. Data are for polycrystalline samples.

tactic was to use a smaller radical anion as the [Fe(C&fe,),]'+ salt on the premise that if spin-pairing bond formation is avoided, it would have a greater spin density which could lead to greater spin interactions.

This philosophy led to the identification of S = l/,[TCNE]'-2 (3). The structure of [Fe(C,Me&]'+[TCNE]'- prepared from acetonitrile contains a molecule of solvent. Upon harvesting of the crystalsll the solvent molecule, however, is completely lost with transformation from a monoclinic into an orthorhombic unit cell. The orthorhombic phase can be directly prepared from tetrahydrofuran. Both phases possess the -*D'+A'-D'+A'--. motif (Figure 3); however, the anion is disordered and unrefinable in the orthorhombic phase.llb All physical measurements were obtained on the orthorhombic phase. The susceptibility of polycrystalline samples of [Fe(C&le5),]'+[TCNE]'- obeys the Curie-Weiss expression, x = C / ( T- e), with 0 = +30 K for T > 60 K (Figures 4 and 6), suggesting dominant ferromagnetic interactions.'l Below 60 K a substantial departure from Curie-Weiss behavior is evident, and below 15 K the magnetization is no longer linearly proportional to the magnetic field (Figure 5 ) . Using a modification of the Faraday technique, i.e., application of a field gradient without an external field other than the Earth's magnetic field, a spontaneous magnetization up to 8 X lo3 (emu G)/mol is observed for polycrystalline samples (Figure 7).llb

Miller et al. I 14-

I

I

I

I

I

-

[Fe(C5Me512). [ T C N E I -

13-

-

12 -

-

Temperature,T. K

T,K

F i g u r e 7. Magnetization (M) as a function of temperature for at several a polycrystalline sample of [Fe111(C5Mes),]'+[TCNE]'magnetic fields. The solid lines are guides for the eye.

F i g u r e 9. Reciprocal magnetic susceptibility as a function of temperature taken for a single crystal of F e ~ ( C 5 M e s ) z l ' + [ T C ~ l ' aligned parallel to the magnetic field. 3 I6

LYn"

-I - D F I T Y

[Fe(C5Me5)2]" [TCNEIIO0

-0 E

\

i 032

a,

X

C l l I i C l l lHponenI y = I 2

[Fe (C5MeS),1'' [ TCNE]'178

oo?OO

316

562

T-T,,

100

K

177

3162

E'

N

I ,

01

ig 2 N4

E

80

3E'

N

I ,

W

Chemical Reviews, 1988, Vol. 88, No. 1 209

Ferromagnetic Charge-Transfer Complexes

t

771 T

E

1.41K

-------

BOOK

I -8

- 6 -4 - 2 0 2 4 6 8 Velocity (mmisec) Relative to Iron

A. Anisotropic Magnetic Properties Preparation of large single crystals facilitating the measurement of magnetic anisotropy has yet to be achieved. Attempts to grow large crystals of the 1-D TCNQ salt are thwarted by the formation of the thermodynamically favored dimer phase, which exhibits independent spin (Curie) magnetic behavior.21@ Large needle crystals of the TCNE salt can be easily grown from acetonitrile; however, the unit cell contains a solvent molecule that is lost upon isolation of the crystals, leading to a polycrystalline sample. Thus, large crystals of the TCNE salt can only be prepared in equilibrium with its supematant." Small single crystals enabling the determination of the magnetic anisotropy of the TCNE salt can be grown from tetrahydrofuran; however, the anion cannot be refined due to disorder.'lb Likewise, disorder of the anion in the [C,(CN),]'- (vide infra) salt, although permitting the determination of the gross structural features, limits detailed microscopic analysis of the structural features important for understanding of the magnetic properties.% Nondisordered large single crystals are a high priority in order to determine the magnetic anisotropy. B. Mossbauer Spectroscopy

The 57FeMossbauer spectra of ferrocene and related compounds have been extensively studied.The Fen compounds, e.g., [Fe11(C5H5)2] [TCNE] exhibit a single, symmetrical quadrupole doublet (Figure 14a) Fem compounds, typical of ferrocene, whereas S = e.g., ( [Fe1n(C5Me5)2]'+)2[TCNQ]22-, exhibit a singlet21b (Figure 14b) typical of ferrocenium. These spectra are essentially invariant with temperature. In an applied magnetic field, a six-line spectrum arising from the external field induced Zeeman splitting of the nuclear levels and corresponding to the usual transition selec,30131

f78K 4

8

VELOClTY (mm Irec) RELATIVE TO IRON

Figure. 15. 67FeMiissbauer spectra as a function of temperature for [Fem(C5Me5)$+[TCNE]'-.

tion rules29dis observed (Figure 14c). In contrast, for the 1:1charge-transfer salts possessing an S = 1/2 anion cation sites, the anions between a pair of planar S = provide an internal dipolar field leading to progressive slow paramagnetic relaxation broadening with decreasing temperature. This, in combination with 3-D ordering processes, results in the observation of Zeeman splitting in zero applied field.9J1p21b*2s*32*33a Thus, as typified by [Fen1(C5Me5),] '+[TCNE]'-, in the absence of an applied magnetic field, six-line Zeeman-split spectra with an internal field of 424 kG (4.3 K) are observed'' (Figure 15). The internal field for this class of compounds varies from 400 to 450 kG (Table 111), which is substantially greater than the expectation of 110 (kG/spin)/Femeassuming only the Fermi contact contribution to internal field. Clearly, a large orbital contribution to the internal field is present. Thus, the observation of unprecedentedly large hyperfine field effects in zero external field Mossbauer spectra is coincident with the identification of a rich variety of magnetic behaviors in the present class of .-.

Miller et al.

210 Chemical Reviews, 1988, Vol. 88, No. 1

1. Alternate Lband Substituents

J 200

t

I

t

1'

t-

600r

E

aoo

1.30R

~

I

t

[

b

I

2.OOK I

l

-10 -9 -8

l

I

I

I

1

/

I

I

I

I

/

I

I

1 2 3 4 5 6 7 8 VELOCITY (mm/sec) RELATIVE TO NATURAL IRON FOIL

9

I

-7 -6 -5 - 4 -3 -2 -1

0

I

I

~ 'L

Figure 16. 57FeMossbauer spectra as a function of temperature [Fe1r1(CSMe5)2]"[TCNQJ*a t 2.0 and 4.3 K.

for

D'+A'-D*+A'-... compounds using more conventional bulk susceptibility methods. The metamagnetic [TCNQ]'- salt exhibits a singlet above 3.6 K (Figure 16a). However, the onset of 3-D ordering occurs below 3.6 K with ultimate observation of a pair of six-line Zeeman-splitspectra2lbin a 2 1 ratio, suggesting an inequivalence of Fe"' sites below 4.2 K (Figure 16b), which is not noted in the 167 K structure determination.21b Heat capacity studies of this system are necessary before the nature of the transition leading to site inequivalence is fully understood. This, in conjunction with the discontinuity in the susceptibility data at 4.0 K, suggests a magnetostructural transformation perhaps accompanying the ordering process or site inequivalence in the magnetic structure of the ordered phase. The Mosebauer spectra for Statal = 3, [Fe"'(C5R5)2]*+[Fe'nX4](R = H, Me; X = C1, Br), which do not possess the --D'+A'-D'+A'--. structural motif, are more complex and are not directly comparable to those materials that belong to the -.D'+A'-D'+A'--- structure type.34g35 C. Structure-Functlon Relatlonshlp

In order to probe the structure-function relationship, with the goal of understanding the structural features necessary to stabilize bulk ferromagnetic behavior of similarly structured, i.e., *-DADADA-, compounds, compounds based on the M(C5R5)2 donors36 were studied. The three synthetic variables that were modified are (1)substitution of the Me groups on the C5ring with H, (2) use of alternate open- and closed-shell, planar and nonplanar anions, and (3) use of other third-row as well as fourth- and fifth-row transitionmetal ions.

I

Ferrocene is more difficult to oxidize (by 0.5 V) than decamethylferrocene and is insufficient to reduce TCNE.37-40 Nevertheless, the ferrocene analogue of [Fe1"(C5Me5)21"[TCNE] *-, Le., [Fe"(C5H5),][TCNE], form^^^^^^' and possess the identical structural motif4' (Figure 3). In contrast to the [Fem(C5Me5)2]'+ salt, this diamagnetic;, green, classical Mulliken charge-transfer complex7 exhibits a broad charge-transfer absorption at -8100 cm-' (-1 eV)39and above 1.6 K a quadrupole doublet 57Fe Mossbauer resonance characteristic of Fe11.29-31These data suggest that complete charge transfer to form stable S = '/2 D's and A's is necessary to achieve stabilization of ferromagnetic behavior. Either a temperature- or pressure-induced "neutralionic" transition8 might be sufficient to lead to the stabilization of ferromagnetic behavior; however, above 2 K a t ambient pressure only Fe'I is observed via Mossbauer spectroscopy (Figure 14a), and no discontinuity is observed in the susceptibility data.30 In principle, more oxidizing acceptors should stabilize complete charge transfer with ferrocene. The chargetransfer [Fe"'(C5H5),]'+[A]'- [A = TCNQF442and C4(CN)6%]complexes were prepared and characterized to possess Fe"' and [A]*- by 57FeMossbauer and infrared spectroscopies. Both materials, however, exhibit weakly antiferromagnetic susceptibilities (i.e., 8 -3 K) and singlet Mossbauer spectra without evidence for zero applied field Zeeman splitting characteristic of ferrocenium. The 1:l TCNQF4 salt does not possess the -D*+A'-D*+A*--. motiP2 observed for the highly magnetic TCNQ,21bp27 TCNE," TCNQ12,32DDQ,33and C4(CN)628charge-transfer salts and does not exhibit cooperative magnetic properties. This points to the importance of this structure type. Since the [Fe"'(C5Me5),]*+salt of [C4(CN),]'- exhibits ferromagnetic interactions28 (vide infra), the structure of the [Fe"I(C5H5),]'+ salt should provide important information on this point. However, crystals suitable for structure determination have yet to be prepared and this issue remains under active research. Likewise, the complex [Fe"'(C5H5)(C5Me5)]*+[TCNE]'has been prepared; however, preliminary susceptibility data do not suggest ferromagnetic behavior.43

-

2. Alternate Afiions

Replacement of [TCNQ]'- with [TCNE]'-,ll [C4(CN)6]*-,%[TCNQ12]'-,32or [DDQ]'-33 leads to similarly structured complexes with dominant ferromagnetic behavior as evidenced by the high-temperature susceptibility obeying the Curie-Weiss expression with 8 2 10 K (Table 111). Replacement of 5' = '/,[TCNE]*with S = 0 [C3(CN)5]-11bleads to formation of a D'+A-D'+A--. structured phase which exhibits essentially Curie susceptibility (e = -1.2 K) with the concomitant lower moment characteristic of highly anisotropic ferrocenes.21bThe nonplanar S = 5/2[Fe111X4](X = C1, Br) Tdanions do not form the --D'+A-D'+A--. structural motif with Fe(C5R5), (R = H, Me). Thus, although they exhibit a variety of interesting complex magnetic behavior,%d5direct comparison to the present system is not appropriate.

..-

Chemical Reviews, 1988, Vol. 88,No. 1 211

Ferromagnetic Charge-Transfer Complexes

ferromagnetism. Ferromagnetic exchange in three dimensions is necessary for bulk ferromagnetic behavior. To achieve ferromagnetic exchange via this mechanism, The S = 0 Co"' analogue of [Ferrr(C5Me5)2J'iroutes to spin pairing, e.g., bond formation, must be [TCNE]'- has been prepared and exhibits essentially avoided. Free radicals, e.g., R',readily dimerize. Stable the Curie susceptibility anticipated for S = 1/2[TCNE]'(0 = -1.0 K).llb Attempts to prepare [Mm(C5Me5)2]'+ free radicals, e.g., [TCNE]'-56 or [TCNQ]*-,21bpossess atoms with unequal positive and negative spin densities (M = Ru, Os) salts of [TCNE]'- have yet to lead to such that it is conceivable that they might form a suitable compounds for comparison with the highly magnetic FeIn phase.& Formation of [ R U " ( C ~ M ~ ~ ) ~ ] * +structure that complies with McConnell's requirement. of the numerous structures reported, none is complicated by disproportionation to R U I ' ( C ~ M ~ ~ )However, ~ have the overlap described by M ~ C o n n e l l .Even ~ ~ with and [ R U ~ " ( C ~ M ~ ~ ) ( C ~ M The ~ ~ COsn1 H ~ analogue )]+.~ these stable radicals dimerization to form diamagnetic has led to the preparation of several phases with TCNE A dimers occurs.27 Examples of C-C a-bond formation and C4(CN),; however, differing stoichiometries, low between C(CNI2moietiesmon adjacent TCNQ's as well susceptibilities, and crystals unsuitable for single-crystal as of more complex chemical reactions59have also been X-ray studies43have hampered progress in this area. reported. Hindered stable radicals, for example, allyl Replacement of Fe"' in [Fe'"(C~Me5)2]'+[A]'- [A = or nitroxide radicals (e.g., qS0and 5,6l respectively), TCNE, C,(CN),] with Ni"I (S = /2) or Cr"' (S = 3/2) might be suitable. However, where evidence exists, they leads to compounds exhibiting cooperative magnetic exhibit Curie or antiferromagnetic behavior. High spin proper tie^.^^ The motivation for studying these comC&NC2H4)6,46*51*92 multiplicity radicals, e.g., = 1 02,62 plexes emanated from the modellgfor the stabilization , ~ ~ ~ ~to~ date ~ ~ ~likewise or S = 4 p o l y c a r b e n e ~ studied of ferromagnetic coupling in molecular solids, and the exhibit either antiferromagnetic coupling or independmagnetic properties of these materials will be discussed ent-spin Curie behavior. in the section devoted to the model. Consequently, available data suggest that the D'+A'-D'+A'-.- structure type with both S 1 1 / 2 D's and S 2 1/2 A's is necessary, but insufficient, for stabilizing cooperative highly magnetic behavior. The key physical properties for [M(C5R5)2]'+[A]'-are summarized in Table 111. 3. Alternate Metals

s

4

V. Models for Ferromagnetic Coupling In Molecular Solids

5

Cleverly designed stable radicals with the proper solid-state packing are sought to test this model. Recently, the spin multiplicity of dicarbenes incorporated Ferromagnetism requires spin alignment throughout into rigid [2.2]cyclophanes has been shown to be conthe bulk, and, though rare for a molecular solid,18the trolled by the overlapping modes of the spins in the quest for molecular-based ferromagnets is the subject aromatic c6 rings.64 McConnell's suggests that of increasing contemporary i n t e r e ~ t . ~ J ~Three l ~ ~ - ~ ~ the singlet pseudometa isomer should possess a singlet mechanisms have been proposed for achieving ferroground state.47*48i64 This is in contrast to the behavior magnetic coupling in a molecular solid and recent reof the quintet pseudoortho and pseudopara isomers of search directed toward this end has started. McConnell the bis(phenylmethyleny1)[2.2]cyclophanes. ESR47@@ proposed two models, namely, Heitler-London spin and computational data%on this system are consistent exchange between positive spin density on a radical and with these results, suggesting that ferromagnetic exnegative spin density on another49and configurational change can be achieved by this McConnell model. admixing of a virtual triplet excited state with the Preliminary results on an intermolecular system are also ground state for a chain of alternating radical cation consistent with McConnell's model;% however, a bulk donors and radical anion acceptor^.^^?^ Mataga52and ferromagnet has yet to be reported. Ovchinnikovs3essentially suggested that very high spin multiplicity alternate hydrocarbon molecules may have ferromagnetic domains. Ferromagnetic superexchange B. High Spin Multlpllclty Molecules and has also been proposed by Soos and co-workerss4and Polymers is discussed with the configurational mixing model for heterospin systems. With the foundation of the ground-state triplet behavior of diphenylcarbene in 1968, Mataga suggested52 that large planar alternate hydrocarbons comprised of A. Heitler-London Spin Exchange meta-substituted triplet diphenylcarbene moieties will In 1963 McConnell stated49that radicals with "...large have a high-spin (ferromagnetically coupled) ground positive and negative atomic a-spin densities ... [that] state via Hund's rule. Bulk ferromagnetic behavior pancake ... so that atoms of positive spin density are requires spin alignment throughout the solid (i.e., interexchanged coupled ... to atoms of negative spin density as well as intramolecular ferromagnetic coupling). A in neighboring molecules ... give a ferromagnetic exmechanism for achieving intermolecular ferromagnetic change interaction". Here ferromagnetic exchange recoupling was not discussed. McConnell's Heitlersults from the incomplete cancellation of antiferroLondon spin-exchange mechanism might suffice; howmagnetic coupled spin components. This is a mechaever, intermolecular ferromagnetic coupling may not be nism for pairwise ferromagnetic exchange,55not bulk necessary if the high spin multiplicity radical was its

Miller et al.

212 Chemical Reviews, 1988, Vol. 88, No. 1

own ferromagnetic domain. E x a m p l e ~ ~ of ~ lhigh ~ f i spin ~ multiplicity polymers that Mataga suggested include 6 and 7.

prior to the realization of an organic ferromagnet via Mataga’s concept include minimization of bond formation prevalent with radicals or spin pairing via distortions and rehybridization; e.g. n

U

Ovchinnik0v5~stated that the spin ( S ) for planar alternate hydrocarbons can be calculated by eq 12,

6

s = In* - 4 / 2

(12)

where n is the number of nonstarred atoms and n* is the number of starred atoms (where adjacent atoms are respectively starred and nonstarred and identically denoted atoms are not adjacent to each other). For example, trimethylenemethanem(9) is predicted to have R 1

7

2

With guidance from McConnell’s Heitler-London spin-exchange model, Iwamura47~48~63*67 and co-workers recently have prepared several high-spin polycarbenes via a route that Mataga suggested, namely, the in situ photolysis of polydiazo derivatives. The best characterized compound is a nonet (S = 4) tetracarbene, 8, prepared via eq 11. The paramagnetic susceptibility

3 9 s

=

1

4 5

11

8

of 8 diluted in either a 2-methyltetrahydrofuran glass or a benzophenone crystal confirms the nonet ground state (peff = 9.08 p~ vs a calculated spin only value of 8.49 pB); however, the temperature dependence of the susceptibility exhibits only independent-spin or antiferromagnetic behavior.63 Preparation of an organic ferromagnet via Mataga’s prescription is problematic. Mataga notes that as the molecule becomes larger, the energy levels form bands and the bondingjhigh-spin nonbonding and nonbonding/antibonding band gaps become smaller. If these gaps become comparable to kBT,then ferromagnetic spin alignment may no longer be feasible as thermal population of higher excited states may occur. Additionally, he states that correlation should lower the energy of the nonmagnetic states, making it more difficult to achieve ferromagnetic behavior. Stable triplet carbenes are required. Carbenes can be further stabilized via greater sp hybridization of the carbene carbon.68 The low-temperature photolytic loss of dinitrogen, eq ll, however, leaves sp2 carbon. Routes to linear sp -C(N,)- carbon, e.g., via addition of steric bulk, should impart greater stability to the triplet. Additionally, chemical problems that must be avoided

s = x/2

S = 1. Likewise, large molecules should have large S’s. The ground state of fused c6 rings (10) are also predicted to have high-spin ground states via the formula S = ( R - 1)/2, where R is the number of r0ws.5~Three and six fused rings ( R = 2 and 3) possess S = and S = 1ground states, respe~tively.~~ Polymers such as 11 are predicted to have a large value for S, Le., x/2!3371 Boron-substituted graphite, BC7 (12), is also predicted to have a high spin state.53 Recently, BC3 has been prepared;72however, other compositions as well as their magnetic properties have yet to be reported.

12 s

= -

Chemical reactivity arising from the unpaired electrons on 10 and 11 ( R = 2 , 3 ) to form bonds suggests that stable high-spin materials are difficult to prepare. Distortions as well as intermolecular coupling which pair up electrons must also be avoided.

Chemical Reviews, 1988, Vol. 88, No. 1 213

FerromagneticCharge-Transfer Complexes D*

(d,2-y2,dHy)

e2g

(6221 '19

9.t

$-

b3g (!'C*)

sf m s = 1/2 ESFO

m E - 1/2 MI= I

Ferromagnetic Coupling

*+ *

ea4

tt9

-k

GSAF

e29

Ms'O

MI= 0

e2g

ttiFSFO

+

9 b39

m s - 1 / 2 mS-

I/2

M E =I

CT Spin Forbidden

b39

ms=o ms-0 ESRFD+R

Antiferromagnetic Coupling

1-

Figure 17. Ferromagnetically (M, = 1) and antiferromagnetically

(M, = 0) coupled paramagnetic ground states.

-t b3g

** -

D* c A'

d3/s1

ms= 1/2 ms= -112

m s = 1 / 2 m s = -1/2 ESRF

fl'

n

Ferromagnetic Coupling

Figure 18. Stabilization of antiferromagneticcoupling via retro charge transfer from an s1 [A]'- to a d3 [D]'+.

Attempts to prepare a polymer similar to 7 based on the iodine oxidation of 1,3,5-triaminobenzenehas been recently reported, although its reproducible chemical or physical characterization has not been d e ~ c r i b e d . ~ ~ Additionally, attempts to prepare a ferromagnetic organic polymer via the formation of a poly(diacety1ene) from a stable nitroxy biradical monomer have been reported.74 In either case magnetization corresponding to only -0.1-29'0 of the maximum theoretical magnetization is realized. The low value and marked variation between samples suggest that the ferromagnetic behavior is not intrinsic.

A? c 0'' b

Figure 19. Schematic state diagram depicting the relative energy

C. Configurational Mixing of an S = 1 Excited State To Stabilize Ferromagnetic Coupling Experimental evidence for ferromagnetic behavior in a molecular solid has been limited to the charge-transfer (2), salt of decamethylferrocene, ( v ~ - C ~ M ~ ~ ) ~with F~I' tetracyanoethylene, TCNE (3).9J1724The mechanisms that govern the stabilization of ferromagnetism in this class of linear-chain alternating radical cation donor (D) and radical anion acceptor (A), i.e., D'+A'-D'+A'-D'+A'-..., molecular charge-transfer complexes are not firmly established; however, it is attractive to considergJlbJgthe admixing of a virtual triplet excited state with the ground-state model originally proposed by M~Connell@'*~~ as this class of complexes possesses both the crystal and electronic structures required by this model. Stabilization of pairwise ferromagnetic coupling (i.e., stabilization of a high-spin, e.g., triplet, state) does not guarantee bulk ferromagnetism in a substance; however, a mechanism to attain bulk ferromagnetism is proposed. The essence of the McConnell model is the configurational interaction of a high-spin excited state to stabilize ferromagnetic coupling. Consider [D]'+ with three electrons in doubly degenerate partially occupied molecular orbitals (POMO), i.e., d3, and an [A]'- with one electron in a nondegenerate POMO, i.e., sl. In the absence of spin interactions there are two ground states for this d3/s1 [D]'+[A]'- pair.75 The Ma = 1ferromag netically coupled (GSFo) and Ma = 0 antiferromagnet-, ically coupled (GSm) states (Figure 17) exist at equal energy and lead to paramagnetic b e h a ~ i 0 r . l ~ As suggested by McConnell the GS lower in energy is the one with the greatest probability of configurational admixing with the lowest energy virtual chargetransfer excited state. Specifically, if an m, = 1excited state (ES) arising from either an m, = 1 donor (D) or an m, = 1acceptor (A), but not both, formed by either virtual (1)retro (Do+ Ao [D]'+ + [A]'-), (2) forward ([D12++ [AI2-- [D]'+ + [A]'-), or (3) disproportionation (e.g., [DI2++ Do 2[D]'+) charge transfer admixes

-

-

of the ground and excited states before and after admixing (a) to stabilize antiferromagnetic coupling for virtual [D]'+[A]' charge transfer and (b) ferromagnetic coupling for virtual [D]'+[A]'charge transfer via the McConnell mechanism. The energy scale is arbitrary.

with the ground state (GS), then the ferromagnetic This coupled ground state (GSFo) will be ~tabi1ized.l~ behavior can be described by using the Hubbard model generalized for the presence of a doubly degenerate leve119924J6(vide infra). For each direction of virtual charge transfer Hund's rule can be used to predict the lowest charge-transfer excited state.19J7Thus, cooperative magnetic coupling [ferro- (FO), ferri- (FI),or antiferromagnetic (AF)] can be predicted for a specific direction of charge transfer. An example is retro charge transfer for the above case (Figure 18). Since the [D]'+ can only accept an m, = -1/2 electron via virtual charge transfer from [A]'- and GSm, not GSFo, has an m, = electron, admixture of the E S A Fexcited ~ ~ ~state will lower the energy of GSM (to GSm) with respect to GSFo and stabilize antiferromagnetic coupling (Figure 19). For virtual forward charge transfer three excited states (ESF~C-D, E&A+D, and EStAF~+-~)are possible (Figure 20). From Hund's rule the easiest excited [D]'+ electron has m, = -1/2 and the lowest excited state is ESFOA+-D. Thus, forward charge transfer for the d3/s1 electron configuration stabilizes the ferromag netically coupled ground state (GSFo) (Figure 19b), which may lead to bulk ferromagnetic behavior as observed24for [Fe"'( C5Me5)2]*+ [TCNE] '-. Since S = 1 [Ferv(C5Me5)2]2+ is an unknown species and S = 0 [TCNEI2- possesses a D2d ground-state geometry,56 the A-D (unlike the D+A) excitation is counterintuitive to chemists. Several solid-state driving forces, however, should stabilize [Fe1V(C,Me5)2]2+[TCNEI2-. They include Madelung stabilization (2+/2should be greater than 0/0)78and Coulombic repulsion (-6 eV for d electrons on the same site79and 2.3 eV for electrons delocalized on a planar organic accep-

-

214 Chemical Reviews, 1988, Vol. 88,No. 1

TABLE 4.npb Magnetic Coupling for Homospin S y ~ t e m s ~ * ' ' ~ D A (or A) (or D) D - A

-

A

S'

S'

AF

Spin AF

d'

S' S'

AF FO

FO AF

S'

AF FO FO AF FO AF FO AF FO FO AF FO

FO AF FO AF FO AF FO AF FO FO AF FO

d3

t' t5 d' d3 t' t5 d3

S'

d' d' d' d'

d3

t'

d3

tj t' t5 tj

d3 t'

t' t5

d2 d2 d2 t2 t* t4

d2 t2 t4 tz t4 t4

AF FO AF FO AF FO

t3

t3

AF

q4

q4

AF

P5

P5

AF

example

D J2

Systems

Spin 1 Systems AF 02,h[ R u ( O E P ) ] ~ ~ ~ AF FO FO AF FO Spin 3 / 2 Systems AF V(C5H,)2'83b Spin 2 Systems AF Spin 5 / 2 Systems AF Mn't(C5H5)2s3a

AF refers to antiferromagnetic coupling and FO to ferromagnetic coupling. bPOMOdegeneracy (intrinsicor accidental): s = singly (a or b), d = doubly (e), t = triply (t), q = quadruply, or p = quintuply. 'Ohmasa, M.; Kinoshita, M.; Sano, M.; Akamatu, H. Bull. Chem. SOC. Jpn. 1968, 41, 1998. dBray, J. W.; Interrante, L. V.; Jacobs, I. S.; Bonner, J. C. Extended linear Chain Compounds; Miller, J. S . , Ed.; Plenum: New York, 1983; Vol. 3, pp 363-415. 'Fischer, E. 0.;Joos, G.; Meer, W. 2. Naturforsch. B 1958, 13b, 456-457. fKarimov, Yu. S.; Chibrikin, V. M.; Shchegolev, I. F. J. Chem. Phys. Sol. 1963, 24, 1683-1685.

tor).56@' Electron transfer predicted to be unfavorable by >0.25 eV frequently occurs in the solid.8' These d3/s1 electron configuration examples lead to contrasting magnetic coupling depending on the direction of virtual charge transfer. Extension of the mechanism to other electron configurations possessing singly (s), doubly (d),triply (t), or higher accidental or intrinsic orbital degeneracies enables the identification of configurations stabilizing ferromagnetic coupling for a specific direction of charge transfer.lg The results of the evaluation of stabilization are summarized in Table 4 for homospin (S, = SA)and Table 5 for heterospin ( S , f SA)systems containing only singly, doubly, and triply degenerate orbitals. For heterospin systems due to incomplete spin cancellation, antiferromagnetic coupling should lead to ferrimagnetic behaviorlg (FI), and the FO and FI interactions can be predicted. 1. Homospin Systems

For an s' [A]'- electron configuration ferromagnetic coupling can be stabilized via forward charge transfer from a D with a d3 electron configuration (Figure 20). As described above the d3/s' [Fem(C5Me5)2]'+[TCNE]'complex possesses a ferromagnetic ground state. For retro charge transfer, although antiferromagnetic cou-

Miller et al.

TABLE

Maenetic CouDlinn for HeterosDin S v ~ t e m s ~ J ' ~

(oPA) ( o r D ) D - A A-D example Spin '/2-Spin 1 Systems S ' d2 FI FIS' FO FI t2 S' FI FO t4 FI FO d' c d2 d' FO FO t2 d' FI FI t4 d3 FO FI d2 d3 FI FI t2 d3 FO FO t4 FI FO t' d2 t' t2 FO FO FI FI t' t4 FO FI t5 d2 FI FI t2 t5 FO FO t5 t4 '

Sl

d' d3 t'

-

Spin 1/2-Spin f Systems FI FI [Crn1(C,Me5)2]'+[TCNE]'-45 FI FO FO FI FI FO FO FI

t3 t3 t3

t5

t3 t3

d2

t3

t2

t3 t3

Spin 1-Spin FI FI FI FO FO FI

do d4 to t6

Spin 1/2-Spin FO FO FO FO

t4

S'

S' S' S*

3/2

Systems

0 Systems

d d d d

aSince SD # SA,FI refers to ferrimagnetic coupling and FO to ferromagnetic coupling. POMO degeneracy (intrinsic or accidental): s = singly (a or b), d = doubly (e), t = triply (t),q = quadruply, or p = quintuply. Proposed for segregated chains of ferromagnetic organic metals.& Proposed for ferromagnetic superexchange (see

re:6"."p;i;ngeilc

1

Rntirerromagnetic Coupling

Aniiferromagnetit Coupling

Figure 20. Stabilization of ferromagnetic coupling via forward charge transfer from a d3 [D]" t o an s1 [A]'-.

pling is predictedlg for the d3/s1 electron configuration, ferromagnetic coupling can result if the D possesses a d' electron configuration. Antiferromagnetic coupling is, however, predicted for d'/s' complexes with forward charge transfer. The [Ni111(C5Me5)2]*+[TCNE]'comp l e possesses ~ ~ ~ the dl/sl electronjc configuration and its susceptibility obeys the Curie-Weiss expression with 8 = -10 K, consistent with dominant antiferromagnetic

interaction^.'^*^^ Virtual disproportionation where an electron on one site is transferred to an adjacent site can stabilize ferromagnetic coupling for a dl/sl or d3/d3electron conf i g ~ r a t i o n '(Figure ~ 21). For example, for the d3/d3 case for ferromagnetically coupled GSFo, the electron

TABLE 6. Electron Configuration of Representative Radicals" electron config

gym-

S

radical

Sl

-

Chemical Reviews, 1988, Vol. 88, No. 1 215

Ferromagnetic Charge-Transfer Complexes

acronym [TCNQ]'-

WQ1'-

metry Dzh C2" C2h

R"

R'

st9

st9

0

st+

stt

m s = 112

m s = -112

MI = 0

MI = 1 ('er;mump~ylic

R'

n

m s = 1 1 2 m s = 112

I

Antiferromagnetic Coupling

+

D3h

[TTF]'+

R'

d3/d3

+

D2h

C2h [TMPD]'+ [MPc]'+

D6

R2*

D2h

-f-f

**

mS=1

m,=o

Ds CZ" D4h D3h

D4h D4h [TCNE]'-

D2h

d'

d2 c 3h C3h

d3

t3

t4

t5

q5 "See ref I9 for references.

with m, = -1/2 can be virtually transferred to the adjacent site which can only accept an electron with m, = -1/2. For the antiferromagnetically coupled GSm the site can only accept an m, = +1/2 electron; however, loss of such an electron from the donor would result in the formation of a higher excited state (Figure 19). Thus, ESFo is lower in energy than ESAFand ferromagnetic coupling is stabilized. This mechanism is invoked between [Fe1"(C5Me5),]*+% as an additional'mechanism for stabilizing ferromagnetic coupling between in-registry chains and provides a means for establishing bulk ferromagnetic behavior as observed for [Fe"'(C5Me5)2J'+[TCNE]'-.11bCobaltocene and its analogues (d') might be a model system to investigate this point as ferromagnetic coupling is predicted for this homomolecular compound. The tl/dl and t5/d3configurations should stabilize ferromagnetic coupling regardless of the direction of virtual charge transfer. Examples of such systems, however, have yet to be identified.

Rz*

Ro

Ms- I

ESFo

ESFO < ESAF

Ro

-f+

**

ml= o

ms= o

ESAF

MI= 0

Figure 21. Schematic illustration of stabilization of ferromagnetic coupling via disproportionation between d3 radicals, R". ESpo is sbbilized as ESAFis a higher excited state than ES,; i.e., ESm < ESm

Assuming the virtual charge transfer involves only the highest energy POM0,84all half-filled POMO cases are expected to exhibit antiferromagnetic coupling. For example, the complex [Ru"(OEP)], (OEP = octaethylporphyrin) possesses parallel chains of groundstate S = 1 (d2) dimers82ain the solid. Preliminary susceptibility data are consistent with the presence of antiferromagnetic beliavior.82b 2. Heterospin Systems

Heterospin systems, i.e., systems with different spin magnitudes on the donors and acceptors, provide an opportunity to obtain ferrimagnetic as well as ferromagnetic solids. Within this simple nearest-neighbor model antiferromagnetic coupling of adjacent spins of the D and A sublattices could only produce incomplete cancellation of the total spin and thus lead to ferrimagnetic behavior. Ferromagnetic superexchange has been proposed49for sl/do (or s1/d4) systems where a virtual triplet is formed via a simultaneous pair of A+D (or D+A) virtual excitations. Attempts to experimentally realize this mechanism with the .-D'+A-D'+A--. complex N,N,N',N'-tetramethyl-p-phenylenediamine pentacyanocyclopentadienide, [TMPD]'+[C,(CN),I-, have not been successful.49 m8D = 1/2; m? = 1. For the lower symmetry s and d electron configurations two combinations support ferromagnetic coupling, i.e., d3/d2with forward charge transfer and d1/d2 with retro charge transfer. Several combinations (d1/t2, d3/t4, t1/t2,and t6/t4)are ferromagnetic invariant of direction of electron transfer. Illustrative systems have yet to be identified for these electron configurations. m f = 1/2; m? = 3/2. Several electron configurations, depending on the direction of virtual charge transfer, can stabilize ferromagnetic coupling. The t3/s1 configuration expected for [Cr(C5Me5)2]+[TCNE]'-due to an accidental degeneracy of the cation e2gand alg orbital~ is ~predicted ~ to exhibit ferrimagnetic coupling for either retro or forward charge transfer. Preliminary magnetic susceptibility data show high susceptibility that is characteristic of either ferri- or ferromagnetic, but not paramagnetic, behavior.45 Detailed analysis is required to distinguish between these two magnetic states.

216 Chemical Reviews, 1988, Vol. 88, No. 1

Miller et al.

msD = 1/2; msA= 1. This electron configuration has been proposed by Dormann, Wudl, and co-workerss4for the design of a ferromagnetic organic metal. They suggest that segregated chains of mixed-valent donors (i.e., S = 1D” and S = ‘/z D*+) with a diamagnetic Acounterion. Examples of suitable materials tQ test this situation have not been reported. D. Generalized Hubbard Model

The importance of the excited state configurational mixing to the lowering of the total ground-state energy in principle can be evaluated by a generalized Hubbard This model is useful in examining the competition between delocalization (i.e., as indicated by the nearest-neighbor charge-transfer integral, /3) and the energy of transferring an electron between the neighboring sites, AE. For AE >> @,the energy difference between singlet (antiferromagnetic) and triplet (ferromagnetic) alignments is CJijAi.Sj,where Siand Sj are the spins on adjacent sites i and j , and Jij is an effective exchange interaction. From perturbation’ analysis the effective exchange interaction, i.e., Jij p 2 / A E , results from each virtual excitation. The sign of each of the contributions of Jij is determined by the spin of the virtual excited state. That is, Jij is negative (ferromagnetic) if the virtual excited state is a triplet and positive (antiferromagnetic) if it is a singlet. The sum of all of the possible exchange interactions between a pair of neighboring sites is Jij C/3,2/AE,,where n refers to each allowed virtual excitation (e.g., as shown in Figures 17 and 18, for d3 D’+ and d’ A*-). Thus, the resulting Jij is the sum of the contributions from each possible excitation weighted by the square of the overlap integral (which varies with excitation and is inversely proportional to the excitation energy) and in principle can be evaluated. The magnetic coupling of the solid is dependent upon the total sum of pairwise interactions among the spins, CJijS&. Hence, the generalized Hubbard model can be used as a guide to the chemical and physical modifications of the solid to enhance the ferromagnetic coupling. For example, chemical modifications (e.g., design features) of the molecular units to decrease AE, and/or increase p, should increase the ferromagnetic exchange interaction and lead to the stabilization of ferromagnetic coupling. Thus, application of pressure should increase the overlap and consequently the relative and absolute magnitude of the /3, and should improve the magnetic properties. Application of this generalized Hubbard model to the permethylmetallocenium radical anion salts is in progress.

-

-

a

b

C

Figure 22. Schematic illustration of (a) mixing of the GSm with m, = 1ESm to lower the energy to GS’po, (b)mixing of the GSm with a pair of m, = 1ESFO(Sarising from intrachain interactions to lower the energy to GS”F0, and (c) mixing of the GSFOwith additional m, = 1ESF$s arising from interchain interactions to further lower the energy to GS”’F0, which may lead to the macroscopic spin alignment necessary for bulk ferromagnetic behavior.

effects (e.g., retro vs forward virtual charge transfer) or magnitude of the stabilization (e.g., inversely proportional to distance and energy difference between the mixing states) may obscure the effect and lead to other phenomena, e.g., para-, meta-, or ferrimagnetism.14 Additionally, other for molecularbased ferromagnetic behavior may be operative. For radicals with intrinsic POMO degeneracies the relatively high symmetry restrids the choice of radicals to specific symmetry point groups. For a d system the molecule or ion must belong to the D2d,Cs, or higher point groups. The orbital symmetry can be accidental. For an intrinsically degenerate t system cubic or icosahedral molecules are necessary. In contrast to the McConnell model’p51 ions may not be necessary, but stable radicals are required. Homomolecular species (D = A) in principle are sufficient if virtual disproportionation ([DI2+ + Do 2 ID]’’) dominates and one of species formed via disproportionation has S > 1/2;vide infra. Since the key point is mixing of an excited state with a ground state with spin conservation, a chain structure as proposed by McConnellWmay not be requisite, but designing materials with strong state mixing is important. Organic or inorganic-based polymer chain and network struct u r e (albeit ~ ~ ~ not molecular solidd8)as well as nonchain structures with the proper admixture of excited and ground states should suffice.

-

E. General Considerations

Assuming the virtual charge transfer involves only the highest energy POM0,85stabilization of ferromagnetic coupling via the McConnell mechanism requires that the stable radicals possess a non-half-filled degenerate POMO. These radicals must not have structural/ electronic distortions that lower the symmetry and break the degeneracies, e.g., the Jahn-Teller effect; however, accidentally degenerate systems (e.g., highspin transition, lanthanide, and actinide metal coordination complexes) suffice. Furthermore, opposing

F. Stabilization of Bulk Ferromagnetism

The admixing of a virtual triplet excited state with the ground state for stabilization of the ferro- or antiferromagnetic coupling model is limited to the repeat Bulk ferrounit, i.e., [Fe*11(C5Me5)2]’+[TCNE]’-. magnetism requires inter- and intrachain spin alignment.llbJ9 Reiterating, mixing of the ESFo with GSFo leads to lowering of the energy to GS’Fo (Figures 19b and 22a). Since the cation is essentially equidistant to a [TCNE]’- above and below it within a chain, virtual

Ferromagnetic Charge-Transfer Complexes [Fe(C5Me5I2]* [TCNE]' ORTHORHOMBIC PHASE

- * INTRACHAIN Fe-Fe = 10621 (216

Figure 23. Unique chains, I, 11, and 111, of orthorhombic

[Fem(C6Me6)2]'+[TCNE]'-.

transfer of an e%.electron forming the admixable M,= 1excited state w t h either [TCNE]' may occur. Thus, two excited states can mix with the GSFo to further lower the energy to GS"FO1lb(Figure 22b), leading to intrachain spin alignment. Intrachain spin alignment leads to energy lowering; however, even with complete intrachain spin alignment (Le., ferromagnetically coupled), if spins on adjacent chains are aligned in the opposite sense, then bulk antiferromagnetic coupling should dominate. Macroscopic ferromagnetism w ill not occur unless ferromagnetic interchain spin alignment occurs. If adjacent chains are out-of-registry by one-half the chain axis length, then [TCNE]'-'s residing in adjacent chains may be comparably separated from a Fem site as are the intrachain [TCNE]'-'s. Thus, the ESP on adjacent chains can additionally mix with GS"F0112 (Figure 22c) to further lower the energy of the system to ES"IFO,leading to the spin alignment throughout the bulk necessary for bulk ferromagnetism. For in-registry chains virtual disproportionation between interchain Fem sites (vide supra) may provide an additional mechanism to align the spins throughout the bulk and contribute to the stabilization of bulk ferromagnetism.llb In contrast, in-registry chains with sl/sl interactions (e.g., [TCNE]'-/ [TCNE]'-) will lead to antiferromagnetic coupling between the chains. The relative magnitude of the resulting interchain ferro- and antiferromagnetic exchange terms will determine if a particular salt will form a 3-D ordered ground state. The unit cell of orthorhombic [Fe111(C5Me5)2]*+[TCNE]'- has three unique adjacent chains, I, 11, and I11 (Figure 23). These in-registry and out-of-registry arrangements (Figure 24) can lead to ferromagnetic and antiferromagnetic as well as unusual magnetic phenomena. V I . Organic Ferromagnets

The magnetic data on [Fe(C5Me5)2]"[A]'- demonstrate that ferromagnetism is achievable in organicbased molecular systems. Replacement of S = 1/2 [A]*-

Chemical Reviews, 1988, Vol. 88, No. 1 217

with S = 0 [A]- demonstrates that the organic [A]'- is a critical ingredient for achieving bulk ferromagnetism. In addition to requiring an unpaired electron in a totally organic p-based orbital, this system, of course, contains iron; however, these charge-transfer complexes are more akin to organic compounds than an inorganic network solid. The iron is low-spin FeTn,not high-spin Fe" or Fe'" or iron metal. Unlike highly magnetic inorganic materials, these complexes are soluble in conventional polar organic solvents. Furthermore, ferrocenes exhibit a chemical reactivity pattern more similar to that of aromatic organic compounds, e.g., benzene,%than that of iron metal or iron oxides. Nonetheless, the quest for an s or p orbital (organic) based ferromagnet maintains academic interest.17i46i48,49,73,74,&1 Relatively high symmetry or an accidental orbital degeneracy is necessary for the stabilization of ferromagnetic coupling by the McConnell mechanism. These situations are documented for transition-metal-based, coordination-based compounds but are rare for organic compounds. Thus, attention focuses toward stable DU,C3,or higher symmetry S = l/z radicals with a degenerate POM0.87 Bres10@*~~ previously pointed out the necessity of a triplet statea7 and has focused research toward the synthesis of stable organic tripletP@ with C3or higher symmetry. His group has successfully prepared S L 1/2 organic radicals, however, neither ferromagnetic coupling nor bulk ferromagnetic behavior has been reported. Radicals possessing D2d symmetry, though rare, in principle may possess a doubly degenerate POMO and should be studied.lg Hexacyanotrimethylenecyclopropane is predicted to have a d4HOMO and oxidation should lead to a d3 radical cation.a9 Since hexacyanotrimethylenecyclopropanehas never been isolated, other derivatives with electron-donating groups, e.g., NMe2, to stabilize the radical cation, e.g., 13, would have to be

7 14

prepared and studied. Alternatively, with the goal of building a solid where adjacent chains are out-of-registry by one-half of a unit cell (as noted for the [Fe(C5Me5)J+[A]'- system) permeta-substituted multilayered c y c l ~ p h a n e se.g., , ~ ~[2.2.2] ~ ~ (14) or [3.3.3], are challenging targets to prepare. These bulky materials may provide the proper solid-state structure enabling bulk ferromagnetic behavior. V I I . Summary Quantitative bulk ferromagnetic behavior (spontaneous magnetization) has been established for the organic-like molecular solid [Fe111(C5Me5)2]'+[TCNE]'-. This complex exhibits magnetic behavior characterized by a Curie-Weiss constant (0)of +30 K, a Curie temperature (T,) of 4.8 K, a saturation magnetization of 16 700 (emu G)/mol, and a coercive field of 1 kG at 2 K. Above 16 K the dominant magnetic interactions are

Miller et al.

218 Chemical Reviews, 1988, Vol. 88, No. 1

E -II

T

64

-

\9 618

-

__

10621

I

I

n

I

m

Figure 24. In-registry and out-of-registry chain interactions for orthorhombic [Fe"1(CSMeS)2]'+[TCNE]'-.11b

along a chain (1-D) and near T,3-D bulk effects as evidenced by the value of the critical exponents dominate the susceptibility. Additionally, six-line zero applied magnetic field Zeeman-split 57Fe Mossbauer spectra with large internal fields (>400 kG) are observed. The structural requirements necessary to prepare molecular ferromagnetic compounds are evolving. The only structure type reported to date possesses parallel -.D'+A'-D'+A'--. linear chains where both the donor (D) and acceptor (A) are radicals. The best studied systems possess decamethylmetallocenium donors; however, replacement of the C5Me5ligand with C5H5does not lead to materials exhibiting ferromagnetic behavior. Preliminary data suggest that c6 ligands can be utilized to prepare charge-transfer complexes exhibiting ferromagnetic behavior. Substitution of [TCNE]'- with [OD&]*-,[C4(CN)6]*-, or [TCNQ12]'- leads to similarly structured complexes exhibiting dominant ferromagnetic behavior. In contrast, the [TCNQ]'- salt exhibits metamagnetic behavior with a Nee1 temperature of 2.55 K and critical field of -1.6 kG. The planar salts (M = Ni, Pt), [TCNQF,]'- salts, [M(S2C2(CN)2]and three polymorphs of the 1:l [C6(CN)6]'-salts as well as the tetrahedral [FeX4]- (X = C1, Br) salts possess different structural motifs and do not exhibit ferromagnetic behavior. Alternate first-row (3d) metallocenes as the [TCNE]'- salt lead to differing magnetic behaviors; i.e., S = 0 Coin is paramagnetic, S = 1/2 Ninl is antiferromagnetic, and S = 3 / 2 Cr"' is ferrimagnetic. Analogous complexes based on second-row (4d) and third-row (5d) Ru"' and Os"'. donors have yet to be prepared. The extended McConnell model was developed and provides the synthetic chemist with guidance for making new molecular materials (organic, organometallic, main group, polymer, and/or inorganic coordination complex) to study cooperative magnetic coupling in systems. Assuming the electron-transfer excitation arises from the POMO, to achieve ferromagnetic coupling in a molecular solid via the McConnell mechanism

a stable radical (neutral, cations/anions, or ions with small diamagnetic counterions) must possess a degenerate POMO that is not half-filled and the lowest excited state formed via virtual charge transfer (retro or forward) possesses the same spin multiplicity and mixes with the ground state. This requirement limits the structure of a radical to D2d,Cs,or higher symmetry where symmetry-breaking distortions do not occur. Intrinsic doubly and triply degenerate orbitals are not necessary and accidental degeneracies suffice. To achieve bulk ferromagnetism, ferromagnetic coupling must be established throughout the solid, and a microscopic model based on the intra- and interchain configurational mixing of excited states was discussed. These requirements are met by [Fe111(C5Me5)2]'+[TCNE]'-, [Fem(C5Me5)2] '+[C4(CN)J*-, etc. Additionally, this model suggests that the Ni"' and Crm analogues should be respectively antiferromagnetic and ferrimagnetic, as preliminary data suggest. The bulk nature of the magnetic behavior emphasizes the importance of not only the primary and secondary, but the tertiary structure in this class of materials. We are probing these effects via acceptor substitution and studying the structure/property relationship. We are looking toward the development of computational methods to aid in the understanding of the structure/magnetic behavior. Extensive chemical syntheses of cleverly designed radicals as well as physical, experimental, and theoretical insight are necessary to test these concepts and establish a deeper understanding of cooperative phenomena in molecular solids. V II I . Acknowledgment

A.J.E. and J.S.M. gratefully acknowledge partial support by the Department of Energy, Division of Materials Science (Grant No. DE-FGOB86ER45271.AOOO). W.M.R. gratefully acknowledges support by the NSF DMR Solid State Chemistry Program (Grant No. 8313710). We thank our co-workers (R. W. Bigelow, J. C. Calabrese, G. A. Candela, S. R.

Chemical Reviews, 1988,Vol. 88,No. 1 219

Ferromagnetic Charge-Transfer Complexes

Chittapeddi, A. Chackraborty, K. R. Cromack, D. A. Dixon, P. J. Krusic, D. M. O’Hare, M. J. Rice, H. Rommelmann, L. Swartzendruber, C. Vazquez, M. D. Ward, D. Wipf, and J. H. Zhang) for important contributions they have made to the success of the work reported herein.

A molecular solid as a solid comprised of low molecular weight molecules or ions that do not possess extended covalent bonding in the solid state. However, shorter than van der Waals separations may be present in the solid. Dissolution into conventional aqueous or organic solvents leads to solvation of the individual ions or molecules that were used to prepare the molecular s01id.l~ Miller, J. S.; Epstein, A. J. J . Am. Chem. SOC.1987, 109, 3850-3856. ... .... . Bechgaard, K.; Jerome, D. Sci. Am. 1982, 247(2), 52-62. Williams, J. M.; Carneiro, K. Adu. Znorg. Chem. Radiochem. 1985,29,249-296.Chaikin, P.M.; Greene, R. L. Phys. Today 1986,31(5),24-32. Jerome, D.; Bechgaard, K. Contemp. Phys. 1982,23,583-624. Williams, J. M. B o g . Znorg. Chem. 1985, 33,183-220. (21) (a) Miller, J. S.; Reis, A. H., Jr.; Gebert, E.; Ritsko, J. J.; Salaneck, w. R.; Kovnat, L.; Cape, T. w.; Van Duyne, R. P. J . Am. Chem. SOC.1979,101,7111.(b) Miller, J. S.; Reiff, W. M.; Zhang, J. H.; Preston, L. D.; Reis, A. H., Jr.; Gebert, E.; Extine, M.; Troup, J.; Dixon, D. A,; Epstein, A. J.; Ward, M. D. J . Phys. Chem. 1987,91,4344-4360. (22) Morrison, W. H., Jr.; Krugsrud, S.; Hendrickson, D. N. Znorg. Chem. 1973,12,1998-2004.Hendrickson, D. N.; Sohn, Y. S.; Gray, H. B. Znorg. Chem. 1971,10,1559-1563. Stryjewski, E.; Giordano, N. Adu. Phys. 1977,26,487. Chittapeddi, S. R.; Cromack, K. R.; Miller, J. S.; Epstein, A. J. Phvs. Rev. Lett. 1987.58. 2695-2698. (a) DLgan, D. M.; HendricGon, D. N. Znorg. Chem. 1975,14, 955-970. (b) Reiger, P. H.; Bernal, I.; Fraenkel, G. K. J . Am. Chem. SOC.1961,83,3918. Baker, G. A.; Rushbrok, G. S.: Gilbert. H. E. Phvs. Rev. A 1964,135,1272. Reis, A. H., Jr.; Preston, L. D.; Williams, J. M.; Peterson, S. W.; Candela, G. A.; Swartzendruber, L. J.; Miller, J. S. J . Am. Chem. SOC.1979,101,2756. Miller, J. S.;Zhang, J. H.; Reiff, W. M. J . Am. Chem. SOC. 1987,109,5484. (a) Bagus, P. S.; Walgren, U. I.; Ahmlof, J. J . Chem. Phys. 1976,64,2324. Ijima, S.;Motoyama, I.; Sano, H. Bull. Chem. SOC.Jpn. 1980,53,3180. (b) Saida, R.; Motoyama, I.; Sano, H. Chem. SOC.Jpn. 1981,106, 6683-6689. (c) Ernst, R. D.; Wilson, D. R.; Herber, R. H. J. Am. Chem. SOC.1984, 106, 1646-1650. (d) Greenwood. N. N.: Gibbs. T. C. Mossbauer Spectroscopy; ‘Chapman and Hall: London, UK, 1971. (e) Zbid., p 103. Miller, J. S.;Zhang, J. H.; Reiff, W. M., in preparation. While Zeeman splitting is not observed for [Fen(CSH,),][TCNE], broadening effects are evident in the doublet spectra (Figure 14a),suggesting other phase transformation behaviors at low temperature. Chittapeddi, S. R.; Epstein, A. J.; Miller, J. S.; Zhang, J.; Reiff, W. M., manuscript in preparation. (a) Miller, J. S.; Krusic, P. J.; Dixon, D. A.; Reiff, W. M.; Zhang, J. H.; Anderson, E.; Epstein, A. J. J. Am. Chem. SOC. 1986,108,4459-4466. (b) Gebert, E.; Reis, A. H., Jr.; Miller, J. S.; Rommelmann, H.; Epstein, A. J. J . Am. Chem. SOC.1982, 104,4403. (34) (a) Zhang, J. H.; Reiff, W. M.; Miller, J. S.; Calabrese, J. C., in preparation. (b) Zhang, - J. H.: Reiff, W. M.: Miller,, J. S.., in preparation. (a) Zhang, J. H.; Reiff, W. M., unpublished results. (b) Reiff, W. M.; Zhang, J. H.: Miller. J. S.. in DreDaratinn. Studies using compll also in progress. Robbins, J. L.; Edelstein, N.; Spencer, B.; Smart, J. C. J . Am. Chem. SOC.1982.104.1882-1893. Webster, 0. W.; Mahler, W.; Benson, R. E. J . Am. Chem. SOC. 1962,84,3678-3684. Rosenblum, M.; Fish, R. W.; Bennett, C. J . Am. Chem. SOC. 1964,86,5166-5170. Brandon, R. L.; Osipcki, J. H.; Ottenberg, A. J . Org. Chem. 1966,31,1214-1217. Adman, E.; Rosenblum, M.; Sullivan, S.; Margulis, T. N. J . Am. Chem. SOC.1967,89,4540-4542. Foxman, B., private communication. Sullivan, B. W.; Foxman, B. Organometallics 1983,2,187-189. Miller, J. S.; Zhang, J. H.; Reiff, W. M. Znorg. Chem. 1987,26, 600-608. O’Hare, D. M.; Green, J. C.; Chadwick, T. P.; Miller, J. S. Organometallics, in press. Kolle, U.; Grub, J. J . Organomet. Chem. 1985,289, 133-139. Miller, J. S.;Epstein, A. J., in preparation. Breslow, R. Mol. Cryst. Liq. Cryst. 1985,125,261. Iwamura, H.; Sugawara, T.; Itoh, K.; Takai, K. Mol. Cryst. Liq. Cryst. 1985,125,251. Iwamura, H. Pure Appl. Chem. 1986,58,187. McConnell, H. M. J . Chem. Phys. 1963,39,1910. McConnell, H.M. R o c . Robert A. Welch Found. Conf. Chem. Res. 1967,11, 144. ~

I X . References (1) Acker, D. S.;Harder, R. J.; Hertler, W. R.; Mahler, W.; Melby, L. R.; Benson, R. E.; Mochel, W. E. J . Am. Chem. SOC.1960, 82, 6408. Kepler, R. G.; Bierstedt, P. E.; Merrifield, R. E. Phys. Reu. Lett. 1960,5, 503. (2) TCNQ = 7,7,8,8-tetracyanoquino-p-dimethane; TCNE = tetracyanoethylene; TCNQIz = 2,5-diiodo-7,7,8,8-tetracyanoqulno-p-dimethane; DDQ = 2,3-dichloro-5,6-dicyanobenzoquinone. (3) See, for example: Miller, J. S., Ed. Extended Linear Chain Compounds;Plenum: New York, Vola. 1-3. Simon, J.; Andre, J. J. Molecular Semiconductors; Springer Verlag: New York, 1985. (4) For detailed overview, see the proceedings of the recent series of international conferences: Synth. Met. 1987,17-I9(Shirakawa, H., Yamabe, T., Yoshino, K., Eds.); Mol. Cryst. Liq. Cryst. 1985.117-121 IPecile. C.. Zerbi. G.. Bozio. R.. Girlandn.

d., Eds.; Chem. Scr. 1981, 17 (Carneiro, .K., Ed.); Lect’, Notes Phys. 1979,95,96 (Bartsic, S.,Bjelis, A., Cooper, J. R., Leontic, B. A., Eds.); Ann. N.Y. Acad. Sci. 1978,313(Miller, J. S.,Epstein, A. J., Eds.). (5) Epstein, A. J.; Miller, J. S. Sci. Am. 1979,241(4),52-61. Bechgaard, K.; Jerome, D. Sci. Am. 1982,247(2),52-61. (6) Little, W. A. Phys. Reu. 1964,134,1416.Sci. Am. 1965,212(2), 21. (7) Mullikan, R. S.;Person, W. B. Molecular Complexes: A Lecture and Reprint Volume;Wiley: New York, 1969. Soos, Z. G. Annu. Rev. Phys. Chem. 1974,25,121. Mullikan, R. S.J . Phys. Chem. 1952,56,801. (8) Batail, P.; LaPlaca, S. J.; Mayerle, J. J.; Torrance, J. B. J. Am. Chem. SOC.1981,103, 951. Mayerle, J. J.; Torrance, J. B.; Crowley, J. T. Acta Crystallogr., Sect. B 1979,B35, 2988. Kanai, Y.; Tani, M.; Kagoshima, S.; Tokura, Y.; Koda, Y. Synth. Met. 1984/1985,10,157.Metzger, R. M.; Torrance, J. B. J . Am. Chem. SOC.1985,107,117. (9) Miller, J. S.;Epstein, A. J.; Reiff, W. M. Zsr. J . Chem. 1987, 27.363-373. Ckdela, G. A.; Swartzendruber, L.; Miller, J. S.; Rice, M. J. J . Am. Chem. SOC.1979,101, 2755. (a) Miller, J. S.; Calabrese, J. C.; Bigelow, R. W.; Epstein, A. J.; Zhang, R. W.; Reiff, W. M. J. Chem. SOC.,Chem. Commun. 1986,1026-1028. (b) Miller, J. S.; Calabrese, J. C.; Rommelmann, H.; Chittapeddi, S. R.; Zhang, J. H.; Reiff, W. M.; Epstein, A. J. J . Am. Chem. SOC.1987,109,769-781. Bray, J. W.; Hart, H. R.; Interrante, L. V.; Jacobs, I. S.; Watkina, G. D.; Prober, D. E. Phys. Reu. Lett. 1976, 35, 744. Bonner, J. C.; Bray, J. W.; Interrante, L. V.; Jacobs, I. S.; Bonner, J. C. Extended Linear Chain Compounds;Miller, J. S., Ed.; Plenum: New York, 1983;Vol. 3, pp 353-415. This relationship breaks down at high applied fields and low tem~erature.’~ (a) Boudreaux, E. A.; Mulay, L. N. Theory and Applications of Molecular Paramagnetism; Wiley-Interscience, New York, 1976;Chapter 1. (b) Kittel, C. Introduction to Solid State Physics, 5th ed.; Wiley: New York, 1976;Chapters 14 and 15. (c) Carlin, R. L. Magnetochemistry; Springer Verlag: New York, 1986. (d) Handbook of Chemistry and Physics, Chemical Rubber Publishing Co.: Boca Raton, FL, 1983;p E107. (e) Handbook of Chemistry and Physics; Chemical Rubber Publishing Co.: Boca Raton. FL. 1983:D E108. DomG C. Contemp. Phys. 1985,26,*49-72.Kadanoff, L. P.; Gotze, W.; Hamblen, D.; Hecht, R.; Lewis, E. A. S.; Palciauskas, V. V.; Rayl, M.; Swift, J. Reu. Mod. Phys. 1967, 39, 395-431. Fisher, M. E. Reu. Mod. Phys. 1974,46, 597-616. White, R. M.Science (Washington, D.C.) 1985,229, 11-15. Wallace, W. E.J. Less Comm. Met. 1984,100, 85-92. (a) Hatfield, W. E.; Estes, W. E.; Marsh, W. E.; Pickens, M. W.; ter Haar, L. W.; Weller, R. R. Extended Linear Chain Compounds; Miller, J. S., Ed.; Plenum: New York, 1983;Vol. 3, pp 43-142. (b) Willett, R. D.; Gaura, R. M.; Landee, C. P. Extended Linear Chain Compounds; Miller, J. S., Ed.; Plenum: New York, 1983;Vol. 3,pp 143-192. (c) Day, P.Acc. Chem. Res. 1979,12,236. (d) Kahn, 0.Angew. Chem.,Znt. Ed. Engl. 1985,24,834-850.

~

220 Chemical Reviews, 1988, Vol. 88, No. 1

(51) Breslow, R.Pure Appl. Chem. 1982,54,927-938. Breslow, R.; Jaun, B.; Klutz, R. Q.; Xia, C.-Z. Tetrahedron 1982, 38, 863-867. (52) Mataga, N. Theor. Chim. Acta 1968,10,372. (53) Ovchinnikov, A. A. Dokl. Nauk Akad. SSSR 1977, 236, 957-961. Theor. Chim. Acta 1978,47,297-304. (54) Radhakrishnan, T. P.; Soos, Z. G.; Endres, H.; Azevedo, L. J. J. Chem. Phys. 1986,85,1126-1130. (55) Ferromagnetic exchange coupling in a molecular solid is essentially equivalent to stabilization of a high-spin, e.g., tri let, state. This is a necessary but insufficient condition for Iulk ferrom netism. (56) Dixon,%. A.; Miller, J. S. J. Am. Chem. SOC. 1987, 109, 3656-3664. (57) E.g.: Herbstein, F. H. Perspect. Struct. Chem. 1977,B33, 2744-2750. Shibaeva, R.P.; Atovmyan, L. 0. J.Struct. Chem. 1972,13,514-531. (58) Hoffmann, S. K.; Corvan, P. J.; Singh, P.; Sethulekshine, C. N.; 1983,105, Metzger, R. M.; Hatfield, W. E. J. Am. Chem. SOC. 4608. Harms, R. H.; Keller, H. J.; Nothe, D.; Wemer, M. Mol. Cryst. Liq. Cryst. 1981,65, 1979. (59) w e r , J. S.;Calabrese, J. C. J. Chem. SOC.,Chem. Commun., in press. (60)Duffy, W., Jr.; Dubach, J. F.; Pianetta, P. A,; Deek, J. F.; Strandburg, D. L.; Miedema, A. R. J. Chem. Phys. 1972,56, 2555-2561. Yamauchi, J. J. Chem. Phys. 1977,67,2850-2855. (61) Kumano, M.; Ikegama,Y.; Sato, T.; Saito, S. Chem. Phys. Lett. 1976,41, 354-356. Boucher, P.D.; Nechtschein, M.; SaintPaul, M. Phys. Lett. A 1973,42,397-398. (62) Comp. Inorg. Chem. 1973,2,705-709. (63) Sugawara, T.; Murata, S.; Kimura, K.; Iwamura, H. J. Am. Chem. SOC.1985,107,5293-5294.Su awara, T.; Bandow, S.; Kimura, K.; Iwamura, H. J. Am. 8hem. SOC.1984, 106, 6449-6450. Sugawara, T.; Bandow, S.; Kimura, K.; Iwamura, H.; Itoh, K. J. Am. Chem. SOC.1986,108,368-371. (64) Izoka, A.; Murata, S.; Sugawara, T.; Iwamura, H. J. Am. Chem. SOC.1985,107,1786-1787;1987,109,2631-2639. (65) Yamaguchi, K.; Fukui, H.;Fueno, T. Chem. Lett. 1986, 625-628. Yamaguchi, K.; Fueno, T.; Nakasuji, K.; Murata, S. Chem. Lett. 1986.629-637. (66) SugawGa, T.; '&kada, H.;Izuoka, A.; Iwamura, H. J. Am. 1986,108,4272-4278. Chem. SOC. (67) Marata, S.;Sugawara, T.; Iwamura, H. J. Am. Chem. SOC. 1987,109,66-126. (68) Hoffmann. R:Zeii. G. D.: Van Dine, G. W. J. Am. Chem. SOC. 1968,90,1&1499: Hunbphreys, R.'W. R.;Arnold, D. R. Can. ,I.. Chern. 2286-2291. - . .-. ... 1976.66. - . ---(69) Dowd, P. Acc. Chem. Res. 1972,5,242. (70) Clar. E.: KemD, W.; Stewart, D. G. Tetrahedron 1958, 3, 325-333. (71) Klein, D. J.; Nelin, C. J.; Alexander, S.; Matsen, F. A. J. Chem. P h p . 1982, 77,3101-3108. (72) Kouvetakis, J.; Kaner, R. B.; Sattler, M. L.; Bartlett, N. J. Chem. SOC.,Chem. Commun. 1986,1758-1759. (73) Torrance, J. B.;Oostra, S.; Nazuil, A. Synth. Met. 1987,19, 708. (74) Korshak, Yu. V.;Ovchinnikov, A. A.; Shapiro, A. M.; Medvedeva, T. V.; Specktor, V. N. Pismo Zh. Eksp. Teor. Ftz. 1986, 43,304-311. Korshak, Yu. V.;Medvedeva, T. V.; Ovchinnikov, A. A.; Specktor, V. N. Nature (London) 1987,326,370-372. Miller, J. S.;Glatzhofer, D. T.; Calabrese, J. C.; Epstein, A. J. J. Chem. SOC.,Chem. Commun., in press. (75) The ferromagnetic charge-transfer complex [Fe"'I

- - I

Miller et al.

(C5Me5)Jt[TCNE]'- with an e (d,~p,d,)~/b (r*)electron configuration possesses the d3@ electron conaguration. (76) (a) Lyon-Caen, C.; C ot, M. J. Phys. C: Solid State Phys. 1975,8,2091. (b) HuiJrbard, J. Proc. R. SOC.London, A: 1963, 276,338; 1964,281,401. (77) Hund's rule was originally propased for atoms (e.g.: McWeeny, R. Coukron's Valence;Oxford University Press: Oxford, 1979; p 102-103);it was later successfully applied to molecules and Eiradicals (e.g.: Lahti, P. M.; Rossi, A.; Berson, J. A. J. Am. Chem. SOC.1985,107,4362-4363.Seeger, D. E.; Lahti, P. M.; Rossi. A.: Berson. J. A. J. Am. Chem. SOC. 1986, 108, 1251-1265). (78) Ladd, M. C. F. Structure and Bonding in Solid State Chemistry; Ellis Horwood: Chichester, UK, 1979. (79) . . Solomon, E. I., Drivate communication. Dixon, D. A., private communicatioi (80) Epstein, A. J.; Lipari, N. 0.; Sandman, D. J.; Nielsen, P. Phys. Reu. B 1976. . --- .- - . - , 13. - - ,-1569-1579. - - - ~- (81) Wheland, R. C.; Gillson, J. L. J. Am. Chem. SOC.1976,98, 3916-3925. Torrance, J. B. Mol. Cryst. Liq. Cryst. 1985,126, 55-67. (82) (a) Collman, J. P.; Barnes, C. E.; Sweptson, P. N.; Ibers, J. A. J. Am. Chem. SOC.1984, 106, 3500-3510. (b) Miller, J. S.; Johnson, D. C., unpublished results. (83) (a) Robbins, J. L.; Edelstein, N.; Spencer, B.; Smart, J. C. J. Am. Chem. SOC.1982,104,1882-1893.Cauletta, C.; Green, J. R.; Kelly, M. R.; Powell, P.; van Tilborg, J.; Robbins, J.; Smart, J. C. J.Electron Spectrosc. Relat. Phenom. 1980,19,327-353. (b)Evans, S.;Green, M. L. H.; Jewitt, B.; King, G. H.; Orchard, A. F. J. Chem. SOC.,Faraday Trans. 2 1974,70,356-376.(c) Gordon, K. R.; Warren, K. D. Inorg. Chem. 1978,17,987-994. (84) Dormann, E.; Nowak, M. J.; Williams, K. A,; Angus, R. O., Jr.; Wudl, F. J. Am. Chem. SOC.1987,109,2594-2599. (85) Circumstances where virtual excitation from a lower lying filled orbital dominates the admixing exciting state are ?os; sible. For example, [CI(CBMe&]' [TCNE]'- has an s /s ground-state electronic configuration and preliminary susceptibility data can be fit by the Curie-Weiss expression with 8 = +12 K, suggesting dominant ferromagnetic interactions, O'Hare, D. M.; Miller, J. S., unpublished results. (86) Cotton, F. A.; Wilkinson, G. Advanced Inorganic Chemistry, 4th ed.; Wiley: New York, 1980;p 1166. (87) This work suggests that a stable tnplet may not be a necessary component for an organic ferromagnet; a stable doublet with a virtually accessible triplet capable of admixing with round state, as observed for [Fem(C5Me&]'+[TCNE]', shodd suffice. (88) Breslow, R. Mol. Cryst. Liq. Cryst. 1985,125,261-267.Breslow, R.;Maslak, P.; Thomaides, J. S. J. Am. Chem. SOC.1984, 106.6453. (89) F. & - --a, T. J. Am. Chem. S O ~1976,98,610-611. . Fukunaga, T.; GorTon, M. D.; Krusic, P. J. J. Am. Chem. SOC.1976,98, 611-613. (90) Hubert, A. J. J. Chem. SOC.C 1967,13-14. (91) Or the C ring analogue with nominal C5 symmetry. (92) Note adcfeed in roof. The dication of hexaazaoctadecahydrocoronene has gee, reported to possess a triplet ground state.""'" Recent structural and magnetic data, however, show that the solid is diamagnetic. Thus, electronic driven structural distortions are an important consideration in the design of an s/p orbital-based organic ferromagnet (Miller, J. S.; Dixon, D. A.; Calabrese, J. C., submitted for publication).